Why Fluency, Not Just Prompting
Learn the difference between knowing a few AI tricks and being fluent with AI, and get a first look at the four competencies this course is built on: Delegation, Description, Discernment, and Diligence.
Tricks Expire. Fluency Lasts.
Most AI training online is a list of tricks: magic phrases, prompt templates, “10 prompts every marketer needs.” Tricks help, but they have two problems:
- They expire. A trick that works with this month’s model may be pointless next month. Interfaces, features, and model behavior keep changing.
- They don’t tell you what to do when the trick fails. A template gets you started. It doesn’t tell you whether AI should be doing the task at all, whether the output is any good, or who is responsible if it’s wrong.
Think about learning a language. A phrasebook gets you a coffee in a new city. Fluency means you can handle a conversation you’ve never had before: you understand how the language works well enough to adapt. AI fluency is the same idea. It is the ability to work with AI effectively, efficiently, ethically, and safely, in situations no template anticipated.
The Four Competencies
This course is built on the 4D AI Fluency Framework, which breaks fluency into four competencies. Each one answers a question you face every time you use AI:
- 1 Delegation
Should AI do this, and which part?
- 2 Description
How do I explain what I need?
- 3 Discernment
Is what came back any good?
- 4 Diligence
Am I using it responsibly, and do I own the result?
| Competency | In one sentence | What it looks like when it’s missing |
|---|---|---|
| Delegation | Deciding whether, when, and how to bring AI into a piece of work. | Using AI for a task it’s bad at, or handing off the part of the job you most needed to do yourself. |
| Description | Explaining what you want clearly enough that AI can help. | Vague requests, generic results, and lots of “that’s not what I meant.” |
| Discernment | Judging the quality of what AI produces and how it got there. | Pasting confident-sounding output into a report without checking it. |
| Diligence | Taking responsibility for how you use AI and what you do with its output. | Sharing confidential data, hiding AI use when it matters, or blaming “the AI” for a mistake you published. |
These are not steps you do once, in order. In practice they overlap: you describe, discern, describe again, rethink what to delegate, and keep diligence in mind throughout. The rest of this course gives each one its own module.
Common mistake: assuming “getting good at AI” means writing better prompts.
Prompting is part of Description, which is only one of four competencies. A perfectly written prompt for a task you shouldn’t have delegated, with output you never checked, shared in a way that breaks your organization’s rules, is not fluent AI use. It just looks fluent.
How This Course Fits
If you’ve taken AI Fundamentals, you already have a head start. Its CRAFT prompting method is a practical tool for Description, and its lessons on verification and responsible use touch on Discernment and Diligence. This course zooms out and gives you a way to reason about any AI interaction, including ones where the right answer is “don’t use AI here.”
Every lesson includes a Try It exercise you can do in any free AI assistant (Claude, ChatGPT, Gemini, Copilot, or similar). Nothing in this course depends on a particular tool.
- Think of the last three times you used AI (or would have liked to). Write each one down in a sentence.
- For each, answer four questions, one per competency:
- Delegation: Was AI a good fit for this task? Did you hand off the right part?
- Description: Did your first request get you close, or did it take several tries?
- Discernment: How did you check the output? What would have happened if it was wrong?
- Diligence: Was there anything you shouldn’t have shared? Did anyone need to know AI was involved?
- Circle the competency where your answers were weakest. Keep this note: you’ll revisit it at the end of the course.
1. What is the main difference between knowing AI tricks and being AI fluent?
Tricks are tied to specific tools and moments. Fluency is a transferable way of thinking that covers whether to use AI, how to explain what you need, how to judge results, and how to stay responsible.
2. A colleague pastes an AI-generated statistic into a client deck without checking where it came from. Which competency is weakest here?
Discernment is judging whether AI output is accurate and useful. An unchecked statistic in a client deck is a classic discernment gap, with a Diligence problem close behind, since the colleague still owns what they present.
3. Which statement about the four competencies is most accurate?
Real AI work loops: you describe, judge, describe again, sometimes rethink what you're delegating, and stay responsible the whole time.
About the Framework
The 4D AI Fluency Framework was created by Rick Dakan and Joseph Feller with Anthropic, and is released under the CC BY-NC-SA 4.0 license. This course’s lessons, examples, and exercises are original. To read the framework in its authors’ own words, visit aifluencyframework.org.
Up Next
Before digging into the competencies, it helps to see the three different ways people work with AI. Each one changes what good delegation, description, discernment, and diligence look like.